RESEARCH - Mathematical modeling of the cause of TB during TNF blockade

[Guest guest]

Arthritis Rheum. 2008 Apr;58(4):947-52.

Mathematical modeling of the cause of tuberculosis during tumor

necrosis factor blockade.

Wallis RS.

PPD Inc., Washington, DC 20005, US

OBJECTIVE: Tumor necrosis factor (TNF) blockade increases the risk of

tuberculosis (TB). The purpose of this study was to use Markov

modeling to examine the contributions of reactivation of latent

tuberculous infection (LTBI) and the progression of new infection with

Mycobacterium tuberculosis to active TB due to TNF blockade. These 2

pathogenic mechanisms cannot otherwise be readily distinguished.

METHODS: Monte Carlo simulation was used to represent the range of

reported values for the incidence of TB associated with infliximab

(TNF monoclonal antibody) and etanercept (soluble TNF receptor)

therapy. Iterative methods were then used to identify for each pair of

incidence rates the Markov model parameters that most accurately

represented the distribution of time to onset of TB as reported to the

Food and Drug Administration. RESULTS: Modeling revealed an apparent

median monthly rate of reactivation of LTBI by infliximab treatment of

20.8%, which was 12.1 times that with etanercept treatment (P<0.001).

In contrast, both drugs appeared to pose a high risk of progression of

new M tuberculosis infection to active TB. Progression of new

infection appeared to cause nearly half of the etanercept-associated

cases; it became the predominant cause of infliximab-associated cases

only after the first year.

CONCLUSION: Despite sharing a common therapeutic target, infliximab

and etanercept differ markedly in the rates at which they reactivate

LTBI. Confirmation of these findings will require the application of

molecular epidemiologic tools to studies of TB in future biologics

registries. Hidden Markov modeling and Monte Carlo simulation are

powerful tools for revealing otherwise hidden aspects of the

pathogenesis of TB.

PMID: 18383389

http://www.ncbi.nlm.nih.gov/pubmed/18383389

--

Not an MD